what is pommerman
>
Introduction
Come meet our team members!
Jiarong Qiu
CS Master student in Viterbi School, USC. Experienced in Computer Vision, Machine Learning.
Wei Cheng
CS Master student at USC. In Pommerman Fight I am responsible for optimizing reward shape and deploy training in A2C/PPO algorithm with RLlib and rwightman framework.
Sairam Kamal Raj
CS Master student at USC. Contributed in training our agent using several RL algorithms and developing statistical report of its performances and comparing efficiencies of several RL algos
Bin Chen
CS master at USC. Explored the A2C algorithm for the pommerman problem and made the PPT and Video for midterm and final presentation.
Shuwei Shi
CS master at USC. Explored the DQN algorithm for the pommerman problem and setup the website for display.
Training Tools
RLlib
- Built-in Loggers
- PyTorch & TF
- Multiple RL algorithms
- Decoupled Envs & Models
Compute Engine
- GTX 2080 (PC)
- P100 (GCloud)
Customized Model
- Fully Connected model
- CNN with LSTM
- CNN derived from VGG-16
